Pavement Crack Segmentation Algorithm Based on Pulse Coupled Neural Network with Brainstorming Optimization
-
摘要: 为提升裂缝检测的分割精度和鲁棒性，基于头脑风暴优化（brainstorming optimization，BSO）和脉冲耦合神经网络（pulse coupled neural network，PCNN），提出了一种路面裂缝图像分割算法（BSO-PCNN）. 该算法采用最大熵准则作为BSO算法的适应度函数，并依据适应度值决定参与次轮迭代的个体；BSO具有强收敛性，可快速确定最优个体解；结合图像特征，获得PCNN模型的最优参数，将其代入PCNN模型实现对裂缝图像的分割. 试验结果表明：算法可在20次迭代内取得不同类型路面裂缝图像的最大适应值，从而确定最佳分割参数；与Sobel边缘检测算法、PCNN图像分割算法、基于最大熵的遗传算法（genetic algorithm based on the maximun entropy of the histogram，GA-KSW）、基于遗传算法参数优化的PCNN分割算法（genetic algorithm based on the pulse coupled neural network，GA-PCNN）相比，BSO-PCNN算法取得了0.9924的区域一致性与0.0900的区域对比度.Abstract: In order to improve the segmentation accuracy and robustness of crack detection, a new segmentation algorithm for pavement crack image is proposed on the basis of brain storming optimization (BSO) and pulse coupled neural network (PCNN). This method uses the maximum entropy criterion as the fitness function of the BSO algorithm, and then determines the individuals participating in the next iteration according to the fitness value. Since the BSO algorithm has strong convergence, which is able to quickly determine the optimal individual solution. Combining the image features, the optimal parameters of the model are obtained, which can be substituted into the PCNN model to achieve the segmentation of the crack image. The experimental results show that the maximum fitness value of different road crack images can be obtained within 20 iterations, so as to determine the best segmentation parameters. Compared with traditional crack segmentation algorithms like Sobel, PCNN, genetic algorithm based on the maximun entropy of the histogram (GA-KSW), and genetic algorithm based on the pulse coupled neural network (GA-PCNN), the proposed method achieves a regional consistency accuracy of 0.9924 and a regional contrast of 0.9924 and a regional contrast of 0.0900.
-
Key words:
- road engineering /
- cracks /
- image segmentation /
- pulse coupled neural network (PCNN) /
- brainstorming
-
表 1 不同算法对比试验
Table 1. Results Experiments of different algorithms comparison
图像编号 原图 Sobel PCNN GA-KSW GA-PCNN BSO-PCNN 1 
2 
3 
4 
5 
下载: 导出CSV
表 2 不同算法的区域对比度和区域一致性
Table 2. Regional contrast and consistency of different algorithms
图像 区域对比度 区域一致性 Sobel PCNN GA-KSW GA-PCNN BSO-PCNN Sobel PCNN GA-KSW GA-PCNN BSO-PCNN 1 0.0487 0.1967 0.0866 0.0736 0.1295 0.9904 0.9909 0.9907 0.9906 0.9909 2 0.0344 0.0719 0.0887 0.0950 0.0912 0.9878 0.9882 0.9885 0.9885 0.9886 3 0.0124 0.0236 0.0513 0.0354 0.0642 0.9962 0.9963 0.9965 0.9964 0.9965 4 0.0814 0.0578 0.1040 0.0938 0.1100 0.9900 0.9898 0.9913 0.9910 0.9913 5 0.0218 0.0375 0.0500 0.0535 0.0538 0.9943 0.9946 0.9947 0.9945 0.9947 平均值 0.0397 0.0775 0.0761 0.0703 0.0900 0.9917 0.9920 0.9923 0.9922 0.9924
下载: 导出CSV
表 3 算法时间和信噪比比较
Table 3. Comparison of algorithm time and signal-to-noise ratio
图像 算法运行时间/s 信噪比/dB Sobel PCNN GA-KSW GA-PCNN BSO-PCNN Sobel PCNN GA-KSW GA-PCNN BSO-PCNN 1 0.596 7.483 0.413 3.289 3.147 2.3408 2.2683 3.0673 1.6554 3.4219 2 0.702 7.378 0.018 3.215 3.017 1.4490 1.9176 3.5000 4.6284 3.4389 3 0.675 7.424 0.023 3.315 3.046 1.8633 1.6814 3.7504 1.4826 3.8259 4 0.648 7.480 0.028 3.253 2.981 0.7415 0.0920 0.1070 0.0578 0.2767 5 0.682 7.632 0.019 3.251 3.003 0.8994 0.0545 1.2121 5.9184 2.6544 平均值 0.661 7.479 0.100 3.265 3.039 1.4588 1.2028 2.3274 2.7489 2.7236
下载: 导出CSV
-
OLIVEIRA H, CORREIA P L. Automatic road crack segmentation using entropy and image dynamic thresholding[C]//Proceedings of the 17th European Signal Processing Conference. New York: IEEE, 2009: 622-626. FAN X, WU J, SHI P, et al. A novel automatic dam crack detection algorithm based on local-global clustering[J]. Multimed Tools Applications, 2018, 77(22): 26581-26599. 徐威,唐振民,吕建勇. 基于图像显著性的路面裂缝检测[J]. 中国图象图形学报,2013,18(1): 70-77.XU Wei, TANG Zhenmin, LV Jianyong. Pavement crack detection based on image saliency[J]. Journal of Image and Graphics, 2013, 18(1): 70-77. 沙爱民,童峥,高杰. 基于卷积神经网络的路表病害识别与测量[J]. 中国公路学报,2018,31(1): 1-10. doi: 10.3969/j.issn.1001-7372.2018.01.001SHA Aimin, TONG Zheng, GAO Jie. Recognition and measurement of pavement disasters based on convolutional neural networks[J]. China Journal of Highway and Transport, 2018, 31(1): 1-10. doi: 10.3969/j.issn.1001-7372.2018.01.001 宋蓓蓓,韦娜. 基于脉冲耦合神经网络的路面裂缝提取[J]. 长安大学学报(自然科学版),2011,31(5): 33-37.SONG Beibei, WEI Na. Pavement cracks extraction based on pulse coupled neural network[J]. Journal of Chang’an University (Natural Science Edition), 2011, 31(5): 33-37. 张坤华,谭志恒,李斌. 结合粒子群优化和综合评价的脉冲耦合神经网络图像自动分割[J]. 光学精密工程,2018,26(4): 962-970. doi: 10.3788/OPE.20182604.0962ZHANG Kunhua, TAN Zhiheng, LI bin. Automated image segmentation based on pulse coupled neural[J]. Optics and Precision Engineering, 2018, 26(4): 962-970. doi: 10.3788/OPE.20182604.0962 周东国,高潮,郭永彩. 一种参数自适应的简化PCNN图像分割方法[J]. 自动化学报,2014,40(6): 1191-1197.ZHOU Dongguo, GAO Chao, GUO Yongcai. Adaptive simplified PCNN parameter setting for image segmentation[J]. Acta Automatica Sinica, 2014, 40(6): 1191-1197. 赵慧洁,葛文谦,李旭东. 最小误差准则与脉冲耦合神经网络的裂缝检测[J]. 仪器仪表学报,2012,33(3): 637-642. doi: 10.3969/j.issn.0254-3087.2012.03.023ZHAO Huijie, GE Wenqian, LI Xudong. Detection of crack defect based on minimum error and pulse coupled neural networks[J]. Chinese Journal of Scientific Instrument, 2012, 33(3): 637-642. doi: 10.3969/j.issn.0254-3087.2012.03.023 宰柯楠,徐江峰. 基于遗传算法和简化PCNN的裂缝检测方法[J]. 计算机应用研究,2017,34(6): 1885-1888. doi: 10.3969/j.issn.1001-3695.2017.06.065ZAI Kenan, XU Jiangfeng. Method of crack detection based on genetic algorithm and simplified pulse coupled neural network[J]. Application Research of Computers, 2017, 34(6): 1885-1888. doi: 10.3969/j.issn.1001-3695.2017.06.065 杨玉婷,史玉回,夏顺仁. 基于讨论机制的BSO算法[J]. 浙江大学学报(工学版),2013,47(10): 1705-1711.YANG Yuting, SHI Yuhui, XIA Shunren. Discussion mechanism based brain storm optimization algorithm[J]. Journal of Zhejiang University (Engineering Science), 2013, 47(10): 1705-1711. CHENG S, QIN Q, CHEN J, et al. Brain storm optimization algorithm:a review[J]. Artificial Intelligence Review, 2016, 46(4): 445-458. doi: 10.1007/s10462-016-9471-0 CHENG S, SHI Y, QIN Q, et al. Maintaining population diversity in brain storm optimization algorithm[C]//Proceedings of the IEEE Congress on Evolutionary Computation. New York: IEEE, 2014: 3230-3237. MAFTEIU-SCAI L O. A new approach for solving equations systems inspired from brainstorming[J]. International Journal of New Computer Architectures and their Applications, 2015, 5(1): 10-18. doi: 10.17781/P001642 LI J, DUAN H. Simplified brain storm optimization approach to control parameter optimization in F/A-18 automatic carrier landing system[J]. Aerospace Science and Technology, 2015, 42: 187-195. doi: 10.1016/j.ast.2015.01.017 ZHOU H, JIANG M, BEN X. Niche brain storm optimization algorithm for multi-peak function optimization[J]. Advanced Materials Research, 2014, 989/990/991/992/993/994: 1626-1630. doi: 10.4028/www.scientific.net/AMR.989-994.1626 GUO X, WU Y, XIE L, et al. An adaptive brain storm optimization algorithm for multiobjective optimization problems[C]//International Conference in Swarm Intelligence. Cham: Springer, 2015: 365-372. 杨玉婷,段丁娜,张欢,等. 基于改进头脑风暴优化算法的隐马尔可夫模型运动识别[J]. 航天医学与医学工程,2015,28(6): 403-407.YANG Yuting, DUAN Dingna, ZHANG Huan, et al. Motion recognition based on hidden Markov model with improved brain storm optimization[J]. Space Medicine and Medical Engineering, 2015, 28(6): 403-407. LI S, PU F, LI D. An improved edge detection algorithm based on area morphology and maximum entropy[C]//Second International Conference on Innovative Computing, Information and Control. New York: IEEE, 2007: 536-539. 曲仕茹,杨红红. 基于遗传算法参数优化的PCNN红外图像分割[J]. 强激光与粒子束,2015,27(5): 38-43.QU Shiru, YANG Honghong. Infrared image segmentation based on PCNN with genetic algorithm parameter optimization[J]. High Power Laser and Particle Beams, 2015, 27(5): 38-43. -
下载:
点击查看大图
图(1) / 表(3)
计量
- 文章访问数: 353
- HTML全文浏览量: 146
- PDF下载量: 12
- 被引次数: 0